[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Bessel Functions and Thumbscrews
From: |
Robert T. Short |
Subject: |
Re: Bessel Functions and Thumbscrews |
Date: |
Wed, 14 Mar 2012 10:10:22 -0700 |
User-agent: |
Mozilla/5.0 (X11; Linux i686; rv:10.0.2) Gecko/20120216 Thunderbird/10.0.2 |
On 03/11/2012 10:29 AM, PhilipNienhuis wrote:
(Sorry for reacting a bit late)
The A&S Bessel functions are fairly "rough" approximations.
The octave code is pretty rough too, but it is fairly good and fairly
fast. More than good enough for what I am doing. It would be good to
have a version of Bessel functions that is more accurate.
This whole thread makes me wonder about the accuracy of everything,
including sin(), cos(), exp(), etc. As you pointed out, many of these
packages were carefully tuned to a particular word size.
I have already spent way more time on this than I really have available
and I won't be solving all these problems anyway, but this has been a
very educational exercise.
Perhaps this NUMAL library can be used for Octave as well. It seems to live
here these days:
http://www.softwarepreservation.org/projects/ALGOL/applications
and the license seems a bit BSD like.
Any license gurus who can shed a light here?
I am putting this on my to-do list. Don't hold your breath though.
Bob